TspanC8 tetraspanins differentially regulate the cleavage of ADAM10 substrates, Notch activation and ADAM10 membrane compartmentalization

22  Download (0)

Full text

(1)

HAL Id: inserm-01677606

https://www.hal.inserm.fr/inserm-01677606

Submitted on 8 Jan 2018

HAL

is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire

HAL, est

destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

TspanC8 tetraspanins differentially regulate the cleavage of ADAM10 substrates, Notch activation and ADAM10

membrane compartmentalization

Stéphanie Jouannet, Julien Saint-Pol, Laurent Fernandez, Viet Nguyen, Stephanie Charrin, Claude Boucheix, Christel Brou, Pierre-Emmanuel

Milhiet, Eric Rubinstein

To cite this version:

Stéphanie Jouannet, Julien Saint-Pol, Laurent Fernandez, Viet Nguyen, Stephanie Charrin, et al..

TspanC8 tetraspanins differentially regulate the cleavage of ADAM10 substrates, Notch activation

and ADAM10 membrane compartmentalization. Cellular and Molecular Life Sciences, 2016, 73 (9),

pp.1895 - 1915. �10.1007/s00018-015-2111-z�. �inserm-01677606�

(2)

O R I G I N A L A R T I C L E

TspanC8 tetraspanins differentially regulate the cleavage

of ADAM10 substrates, Notch activation and ADAM10 membrane compartmentalization

Ste´phanie Jouannet1,2Julien Saint-Pol1,2Laurent Fernandez3,4

Viet Nguyen2Ste´phanie Charrin1,2 Claude Boucheix1,2 Christel Brou5 Pierre-Emmanuel Milhiet3,4 Eric Rubinstein1,2

Received: 29 June 2015 / Revised: 3 November 2015 / Accepted: 3 December 2015 / Published online: 19 December 2015 The Author(s) 2015. This article is published with open access at Springerlink.com

Abstract The metalloprotease ADAM10 mediates the shedding of the ectodomain of various cell membrane proteins, including APP, the precursor of the amyloid peptide Ab, and Notch receptors following ligand binding.

ADAM10 associates with the members of an evolutionary conserved subgroup of tetraspanins, referred to as TspanC8, which regulate its exit from the endoplasmic reticulum. Here we show that 4 of these TspanC8 (Tspan5, Tspan14, Tspan15 and Tspan33) which positively regulate ADAM10 surface expression levels differentially impact ADAM10-dependent Notch activation and the cleavage of several ADAM10 substrates, including APP, N-cadherin and CD44. Sucrose gradient fractionation, single molecule tracking and quantitative mass-spectrometry analysis of the repertoire of molecules co-immunoprecipitated with Tspan5, Tspan15 and ADAM10 show that these two tet- raspanins differentially regulate ADAM10 membrane

compartmentalization. These data represent a unique example where several tetraspanins differentially regulate the function of a common partner protein through a distinct membrane compartmentalization.

Keywords Membrane compartmentalizationNotch ADAM10 TetraspaninEctodomain shedding Microdomain

Introduction

Members of the ADAM (a disintegrin and metalloprotease domain) family are membrane-anchored metalloproteases that mediate a proteolytic cleavage of various transmem- brane proteins within their extracellular region. This process, referred to as ectodomain shedding, plays an important role in various cell and developmental processes [1, 2]. ADAM10 mediates the ectodomain shedding of more than 40 transmembrane proteins, including cytokine and growth factor precursors, as well as adhesion proteins such as E and N-cadherins [2]. ADAM10-mediated cleavage of the amyloid precursor protein (APP) prevents the formation of the amyloid peptide Ab, a major com- ponent of amyloid plaques observed in Alzheimer’s disease [3]. ADAM10 is also the main protease for the cleavage of Notch receptors at a site called S2 following ligand bind- ing. This step is a prerequisite for a second cleavage at the S3 site by thec-secretase complex that results in the release of Notch intracellular domain (NICD), which translocates to the nucleus and regulates the transcription of Notch target genes [4–7]. Importantly, ADAM10-deficient mice die during the development, and its tissue-specific ablation yields abnormalities in various organs that are associated with a defect in Notch signaling [8–11].

S. Jouannet and J. Saint-Pol contributed equally to this work.

Electronic supplementary material The online version of this article (doi:10.1007/s00018-015-2111-z) contains supplementary material, which is available to authorized users.

& Eric Rubinstein

eric.rubinstein@inserm.fr

1 Inserm, U935, 94807 Villejuif, France

2 Universite´ Paris-Sud, Institut Andre´ Lwoff, 94807 Villejuif, France

3 Inserm, U1054, 34090 Montpellier, France

4 Universite´ de Montpellier, CNRS, UMR5048, Centre de Biochimie Structurale, Montpellier, France

5 Institut Pasteur, Laboratoire ‘‘Signalisation et Pathogene`se’’, 75015 Paris, France

DOI 10.1007/s00018-015-2111-z

Cellular and Molecular Life Sciences

(3)

Ectodomain shedding is a tightly regulated process and has been shown to be stimulated by various stimuli. For example, PKC activators and various GPCR ligands have been shown to strongly stimulate ADAM17-dependent shedding of various substrates, including growth factors and cytokines [1, 12]. The activity of ADAM10 towards various substrates has been shown to be stimulated by calcium ionophores, activation of P2X7 receptor as well as by mAb directed against several members of the tetra- spanin superfamily [12–14]. The differential ability of ADAM10 and ADAM17 to support the proteolysis and activation of normal and mutant forms of Notch also sug- gests a tight regulation of these proteases. Indeed, ADAM10 is essential and ADAM17 dispensable for ligand-dependent Notch activation in cellular models [5–

7]. This is consistent with the lack of Notch-loss of func- tion phenotype in ADAM17-null animals [15]. In contrast, mutant forms of Notch that are active independently from the presence of ligand can be cleaved at the S2 site by other proteases including ADAM17 [5,6].

The expression of ADAM10 and Notch activity are regulated by several members of the tetraspanin super- family [16–19]. Tetraspanins are expressed by all metazoans, and are characterized by four transmembrane domains that flank two extracellular domains of unequal size, conserved key residues, and a specific fold of the large extracellular domain. Genetic studies in humans or mice have shown their key role in a number of physiological processes including reproduction, vision, immunity, kidney function, muscle regeneration and mental capacity [20–22].

A major feature of these molecules is to associate with many other integral proteins, thus organizing a dynamic network of interactions referred to as the ‘‘tetraspanin web’’ or tetraspanin-enriched microdomains [20–22]. The organization of this web has been resolved, at least in part, with tetraspanins interacting directly with a limited number of partner proteins to form primary complexes which in turn associate with one another. Several tetraspanin/partner pairs have been identified. For example, CD151 associates directly with the laminin-binding integrinsa3b1 anda6b1 [23,24], and CD9 and CD81 share two common partners, CD9P-1 and EWI-2, two related Ig domain proteins [25–

28]. Tetraspanins regulate various properties of the mole- cules they associate with, including their trafficking, the binding of ligands, downstream signaling, and for ectoen- zymes, their enzymatic activity [20–22,29].

We and others have recently demonstrated that ADAM10 has six tetraspanin partners, which mediate its exit from the ER and belong to a subgroup of tetraspanins having eight cysteines in the largest of the two extracellular domains and referred to as TspanC8 [16,18,19]. This level of regulation appears to be important for Notch signaling and is evolutionary conserved. Indeed, silencing Tspan5

and Tspan14 reduced Notch activity in a human cell line, in association with a reduction of ADAM10 surface expres- sion [16]. Mutations of the TspanC8 tetraspanin Tsp-12 in Caenorhabditis elegans genetically interacted with Notch or ADAM10 mutations [17]. Finally, depletion of the three DrosophilaTspanC8 tetraspanins impaired several Notch- dependent developmental processes, Notch activity and ADAM10 subcellular localization in vivo [16].

Direct association of ADAM10 with several tetraspanin partners suggests that some of its properties could be reg- ulated differently depending on the tetraspanin with which it is associated. We show here that the TspanC8 tetra- spanins Tspan5, Tspan14, Tspan15 and Tspan33 have a different impact on ADAM10-dependent functions. In particular, Tspan33 and Tspan15 appear to be negative regulators of ligand-induced Notch activity. We also show that Tspan5 or Tspan15 differentially affect the membrane compartmentalization of ADAM10 as shown by confocal microscopy analysis, single molecule tracking and the analysis of their repertoire of co-immunoprecipitated molecules. These data present strong evidence that tetra- spanins can regulate the function of their partner proteins by acting on their membrane compartmentalization.

Results

Tspan15 is a negative regulator of Notch activity

We have previously demonstrated that silencing Tspan5 and Tspan14 in U2OS cells transduced with human Notch1 (U2OS-N1) decreased ADAM10 surface expression levels and Notch activity. We could not test the role of Tspan15 and Tspan33 in these cells which do not express these two tetraspanins. To directly compare the effect of Tspan5, Tspan14, Tspan15 and Tspan33 on Notch activity, we stably expressed these TspanC8 in U2OS-N1 cells. All 4 tetraspanins were expressed at the cell surface as determined by labeling with membrane impermeable biotin (Fig. 1), associated with endogenous ADAM10 and stimulated a 3- to 5-fold increase in ADAM10 surface expression levels. In contrast, there was no change of Notch expression (Fig.1). To exam- ine the impact of the expression of these TspanC8 on ligand-induced Notch activity, the different cell lines were co-cultured with OP9 cells expressing or not the Notch ligand DLL1. The expression of Tspan5 or Tspan14 had no significant effect on Notch activity. In contrast, U2OS-N1 cells expressing Tspan15 or Tspan33 showed a*60 % decrease in OP9-DLL1-induced Notch activity as compared to U2OS-N1 cells (Fig.2a). In addition, cells transfected with Tspan15 and Tspan33 also showed diminished Notch signaling in response to

(4)

immobilized DLL1, indicating that these tetraspanins do not modulate Notch signaling by changing the interac- tion of U2OS-N1 cells with OP9-DLL1 cells (Fig.2b).

In addition, the transfection of Tspan15 or Tspan33 did not change the expression level of endogenous Tspan5 and Tspan14, as determined by RT-qPCR (data not shown). Additional experiments were performed to further characterize the effect of Tspan15 on Notch signaling. The inhibition of Notch signaling is not due to the selection of a sub-population of U2OS-N1 cells having a lower ability to respond to Notch activation because a second independent cell population of cells expressing Tspan15 showed similar decrease in Notch

signaling (Fig. S1). In addition, silencing Tspan15 in U2OS-N1/Tspan15 cells restored Notch signaling (Fig.2c). Tspan15 expression did not reduce the activity of two constitutively active Notch constructs (Fig. 2d):

NICD, which corresponds to the intracellular domain of Notch1 lacking the PEST domain, and Notch1-DE, which contains a short extracellular stub, the trans- membrane domain and the intracellular domain of Notch1 without the PEST domain [30–32]. The activity of both constructs is independent from ADAM10 activity, whereas the activity of Notch1-DE, but not NICD, requires c-secretase activity. Thus, Tspan15 acts at a pre-c-secretase step.

B

IP GFP

70 40 70 40 70 40

(KDa)

Strepta.

ADAM10

(ADAM10+) GFP C

101 102 103

GFP

ADAM10

Tspan15

Tspan5 Tspan14

CD9

100 104

101 102 103

Control

Tspan33

100 104

101 102 103

101 102 103 101 102 103

A

Extracts

GFP ADAM10

tubulin

70 55 100

-

Notch1

55 (KDa)

Transfection: Transfection: -

ADAM10

-

TspanC8

Fig. 1 Expression of four TspanC8 tetraspanins in U2OS-N1 cells.

aFlow-cytometry analysis of the surface expression of ADAM10 in U2OS-N1 cells stably expressing GFP-tagged TspanC8 tetraspanins or CD9. b Western-blot analysis of the expression of Notch1, ADAM10 and tetraspanins in U2OS-N1 cells stably expressing GFP- tagged TspanC8 or CD9.cAfter biotin labeling of surface proteins, U2OS-N1 cells stably expressing or not GFP-tagged Tspan5, Tspan14, Tspan15 and Tspan33 were lysed and the interaction of

these tetraspanins with ADAM10 was analyzed by co-immunopre- cipitation using GFP-trap beads and Western blot. The major 68 kDa band revealed by the anti-ADAM10 mAb perfectly overlapped with the band labeled ADAM10 in the upper panel (black arrowhead).

Immunoblotting with the GFP antibody was done after ADAM10 immunoblotting. All experiments were performed at least three times with similar outcome

(5)

We then investigated whether endogenous Tspan15 has also an inhibitory effect on Notch signaling. PC3 prostate cancer cells have been shown to express Notch-1 at the protein level [33]. RT-qPCR analysis has shown that these cells express mainly Tspan15 and Tspan5 but no Tspan14 [16]. As shown in Fig.3a, Notch activity, determined using the luciferase assay, was 2.5 times higher in cells co-cultured with OP9-DLL1 cells than in cells co-cultured with OP9 cells. This activation is due to canonical Notch activation because it was blocked by DAPT, ac-secretase inhibitor, and by the ADAM10 inhibitor GI254023X (Fig.3a). Silencing Tspan15 in these cells, with two previously validated siRNA, decreased Tspan15 mRNA levels by*90 % [16], decreased

ADAM10 surface expression levels by*60 % (Fig.3c), but increased Notch activity induced by OP9-DLL1 cells two- fold (Fig. 3b). In contrast, silencing Tspan5 had less impact on ADAM10 expression (*20 % reduction in surface expression levels) but reduced Notch activity in response to OP9-DLL1 cells by*50 % (Fig.3b). Tspan5 depletion also prevented the increase in Notch signaling observed upon silencing Tspan15.

Altogether these data indicate that the different TspanC8 have a different impact on Notch signaling. In particular, Tspan15 is a negative regulator of this signaling, acting at a step upstream thec-secretase step, probably by regulating ADAM10 activity.

C

0 50 100 150

Transfection

*** ***

Notch activity(% of control)

Stimulation: OP9+/-DLL1 A

0 50 100 150

***

**

Transfection Stimulation: DLL1-Fc

Notch activity(% of control)

B

0 200 400

Notch activity(% of control)

600

800 U2OS N1/Tspan15

RNAi:

0 100 200

300 U2OS-N1

NICD Notch∆E

U/Tspan5 U/Tspan15

Notch activity(% of control)

D

Fig. 2 Transfection of Tspan15 and Tspan33 reduces Notch activity in U2OS-N1 cells.aNotch activity, measured using a CSL reporter luciferase assay, of U2OS-N1 cells stably expressing or not Tspan5, Tspan14, Tspan15 or Tspan33. Notch was activated by incubation of the cells with OP9-DLL1 cells for 20–24 h. Thegraph shows the mean±SEM of ten (Tspan5, Tspan15) or four (Tspan14, Tspan33) independent experiments in duplicate. In each experiment, the signal obtained is expressed as a percentage of the signal observed for non- transfected U2OS cells. bNotch activity of U2OS-N1 cells stably expressing or not Tspan5, Tspan14, Tspan15 or Tspan33. Notch was activated by incubation of the cells on DLL1-Fc-coated tissue culture plates for 20–24 h. The graph shows the mean±SEM of five

(Tspan5, Tspan15), four (Tspan33) or two (Tspan14) independent experiments in duplicate. Each experiment is normalized on the signal obtained for non-transfected U2OS cells.cU2OS-N1/Tspan15 were treated with a control siRNA or a siRNA targeting Tspan15 before analysis of Notch activity. The graph shows the mean±SEM of three independent experiments performed in duplicate.dU2OS-N1 cells stably expressing or not Tspan5 or Tspan15 were transiently transfected with NICD and Notch1-DE constructs. Notch activity was determined using the luciferase assay. The graph shows the mean±SEM of two independent experiments performed in dupli- cate. ***p\0.001 and **p\0.01 as compared with control cells

(6)

TspanC8 tetraspanins differentially regulate the cleavage of ADAM10 substrates

We then tested whether the expression of the different TspanC8 in U2OS cells modified the cleavage of other well-characterized ADAM10 substrates. For this purpose we analyzed the production of the carboxy-terminal membrane stub (CTF) generated upon cleavage using antibodies to the cytoplasmic domain. To prevent degra- dation of this fragment, cells were treated with DAPT, ac- secretase inhibitor. As shown in Fig.4a, U2OS-N1 cells expressing Tspan15 showed a 80 % reduction in APP CTF production as compared to parental cells. A partial reduc- tion of APP CTF production was also observed for cells expressing Tspan14 or Tspan33, but not for cells express- ing Tspan5.

To examine the role of endogenous Tspan5 and Tspan15, we studied the cleavage of ADAM10 substrates in PC3 cells (Fig.4b, c). In these cells, the production of APP, N-cadherin and CD44 CTF were dependent on ADAM10 activity as shown by the inhibition by the ADAM10 inhibitor GI254023X (Fig.4b, c, top). Tspan15 silencing by three different siRNA reduced the production of N-cadherin CTF by *75 % on average, and slightly increased the production of APP CTF (Fig.4b, c, bottom).

We were unable to conclude about the role of Tspan5 in regulating the cleavage of APP or N-cadherin in this model because the three different siRNA used had a different effect, yielding no change on average. (Fig.4b, c, middle).

In contrast, all three Tspan5 siRNA reduced the production of CD44 CTF (Fig.4b, c, middle; on average the reduction of CD44 CTF production is *55 %). The production of CD44 CTF was not modified upon Tspan15 silencing (Fig.4b, c, bottom).

Altogether, these data indicate that the different TspanC8 tetraspanins have a different impact on the cleavage of several ADAM10 substrates.

Differential membrane compartmentalization of ADAM10 according to the expression of Tspan5 or Tspan15

We reasoned that the differential activity of TspanC8 on Notch activity could be due to a different compartmental- ization of ADAM10. Sucrose gradient fractionation was used to determine the impact of these proteins on the membrane environment of ADAM10. In initial experi- ments, we found that ADAM10 (and the other molecules tested here) was nearly completely solubilized after lysis using Brij97 (data not shown). In contrast, a fraction of C

0 50 100

ADAM10 expression (% of control) siRNA ctrl siRNA Tspan5 #1 siRNA Tspan15 #1 siRNA Tspan15 #2

+- --

+- --

-- +-

-- +-

*** ***

**

***

+- +- A

*** ***

Notch activity (% of control) 0 50 100 150

DAPT:

GI254023X: -

- -

- +

- -

+ OP9/DLL1 OP9

***

B ***

0 50 100 150 200

250 ***

*** **

siRNA ctrl:

siRNA Tspan5 #1:

siRNA Tspan5 #2:

siRNA Tspan15 #1:

siRNA Tspan15 #2:

+- -- -

OP9/DLL1 OP9

+- -- -

+- -- -

-- +- -

-- -- +

+- -- +

***

-- +- - Notch activity (% of control)

Fig. 3 Tspan5 and Tspan15 are positive and negative regulators of Notch activity in PC3 cells.aNotch activity in PC3 cells measured using a CSL reporter luciferase assay. PC3 cells transfected with the reporter construct were incubated with OP9 or OP9-DLL1 cells for 20–24 h. The cells were incubated or not with the ADAM10 inhibitor GI254023X (3lM) or thec-secretase inhibitor DAPT (3lM). The graphshows the mean±SEM of three independent experiments in duplicate.bNotch activity, measured using a CSL reporter luciferase assay, of PC3 cells treated with the indicated siRNA. Notch was activated by incubation with OP9-DLL1 cells, or with OP9 cells as a control for endogenous activity. Thegraphshows the mean±SEM of three independent experiments performed in duplicate. cFlow- cytometric analysis of the surface expression of ADAM10 in PC3 cells treated with the indicated siRNA. The graph shows the mean±SEM of four independent experiments. ***p\0.001 and

**p\0.01 as compared with control cells or other conditions

(7)

(KDa)

130

55 170

70 100

35 40 25 15 10

mature immature

CTF

none Tspan5

DAPT: - + - + - + - + - +

Transfection: Tspan14 Tspan15 Tspan33

Blot:

APP

mature immature

DAPT:

GI254023X:

- -

+ -

+ +

- -

+ -

+ + RNAi: Control Tspan15

(KDa)

130

55 170

70 100

35 40 25 15 CTF 10

Blot:

APP B A

(KDa)

130

55 170

70 100

35 40 25 15 10

CTF N-cadherin Blot:

N-cadherin

- -

+ - Tspan5 -

- + - Control

CTF CD44

(KDa)

130

55 170

70 100

35 40 25 15 10

Blot:

CD44 DAPT:

GI254023X:

- -

+ -

+ +

- -

+ -

+ +

- -

+ -

+ + RNAi: Control Tspan5 Tspan15 DAPT:

GI254023X:

- -

+ -

+ +

- -

+ -

+ + RNAi: Control Tspan15

- -

+ - Tspan5 -

- + - Control

0 50 100 150

APP N-Cad CD44

DAPT DAPT+

GI254023X

*** ***

***

CTF production

C

0 50 100 150

***

Transfection

* ***

CTF production

0 50 100 150

CTF production

si Ctrl siTspan15

***

*

APP N-Cad CD44

0 50 100 150

si Ctrl siTspan5

CTF production

***

APP N-Cad CD44

(8)

ADAM10 partitioned into the light fractions of sucrose gradients after lysis with a milder detergent, Brij98 (Fig.5). Importantly, the fraction of ADAM10 present in low density fractions was higher in cells transfected with Tspan5 than in cells transfected with Tspan15 or in par- ental U2OS-N1 cells, indicating a change in the membrane environment of ADAM10 in these cells. As a control, the partitioning into the different fractions of CD9 and CD9P-

1, a CD9 and CD81 molecular partner, was similar in the different cell lines.

Dynamics and partitioning of ADAM10 in U2OS-N1 cells expressing or not Tspan5 or Tspan15 were next investigated using single-molecule tracking (SMT). In this technique, the labeling of a low number of molecules allows individual molecules to be optically isolated and their position accurately determined. Frame by frame positioning of the proteins allows reconstruction of their trajectories, calculation of their apparent diffusion coeffi- cient (ADC) as well as the determination of their mode of diffusion. Similar to other membrane proteins, including CD9 and CD81 [34–36], the distribution of ADAM10 ADC was large (Fig.6a) and three modes of diffusion were identified (Fig.6b, c): (1) pure Brownian diffusion, (2) pure confined or restricted diffusion and (3) diffusion with different combinations of Brownian and confined modes referred to as ‘‘mixed trajectories’’. The behavior of ADAM10 molecules in cells expressing Tspan15 was dif- ferent from that of parental cells or cells expressing Tspan5. The average ADC of ADAM10 was significantly higher in cells expressing Tspan15 than in parental cells or cells expressing Tspan5 [0.104±0.032 lm2/s versus

bFig. 4 TspanC8 differentially regulate the cleavage of ADAM10 substrates.aWestern-blot analysis of APP in U2OS cells expressing or not the various TspanC8, after incubation for 24 h in DMSO or the c-secretase inhibitor DAPT. The graph on the right shows a quantification of the production of APP CTF (mean±SEM) of three independent experiments.bRepresentative Western-blot anal- ysis of APP (top), N-cadherin (middle) and CD44 (bottom) in PC3 cells treated with a control siRNA or siRNA targeting Tspan5 and Tspan15, and incubated for 24 h with DMSO, DAPT or a combina- tion of DAPT and the ADAM10 inhibitor GI254023X.

cQuantification (mean±SEM) of the effect of GI254023X, and siRNA targeting Tspan5 or Tspan15 on the production of APP, N-cadherin (N-cad) and CD44 CTF. The experiments were performed twice with three different siRNA for each tetraspanin. To take into consideration the potential variability of the effect of different siRNA, the mean and SEM were calculated on the data obtained for all three siRNA

Fractions 1-5 Fractions 6-9

130

25 70 3 4 5 6 7 8 9

Top Bottom

ADAM10

CD9

U2OS/Tspan5 U2OS/Tspan15

CD9P-1

70 GFP

(TspanC8)

Top Bottom

Blot: 3 4 5 6 7 8 9 1 2 1 2 3 4 5 6 7 8 9 (KDa)

Top Bottom

U2OS-N1

1 2

CD9P1

0 20 40 60 80

CD9

0 20 40 60 80 ADAM10

0 20 40 60 80

GFP-TspanC8

0 20 40 60 80

U2OS Tspan5

U2OS Tspan15 U2OS

N1

U2OS Tspan5

U2OS Tspan15

U2OS Tspan5

U2OS Tspan15 U2OS

Tspan5 U2OS Tspan15 U2OS

N1

U2OS N1

% in fractions

*

Fig. 5 Tspan5 and Tspan15 differentially affect ADAM10 solubi- lization. U2OS cells transfected with Tspan5 or Tspan15 were lysed in the presence of Brij98. The lysates were directly subjected to equilibrium density gradient centrifugation. Gradient fractions were collected and analyzed by Western blot as indicated. Thegraphsshow

the relative abundance (mean±SEM) of the indicated proteins in low (fractions1–5) and high (fractions6–9) density fractions in three independent experiments. *p\0.05 as compared with the other cell types

(9)

Confined Mixed Brownian

Distribution of ADAM10 trajectories (%)

Tspan5 0

20 40 60 80 100

16 69

23 56

Tspan15 Cell type WT

53

C

Tspan15 Tspan5 WT

Tspan15 Tspan5 WT Brownian

Confined

Mixed Tspan15

Tspan5 WT

0.112 ±0.030 0.108 ±0.027 0.138 ±0.030 0.030 ±0.012 0.032 ±0.013 0.041 ±0.016 0.007 ±0.004 0.010 ±0.006 0.009 ±0.006

B ADC (μm2/s)

0.001 0.01 0.1 1

0.001 0.01 0.1 1

Tspan15 Tspan5 All trajectories

WT 0.067 ±0.027

0.070 ±0.026

0.104 ±0.032

GFP ADAM10 ADAM10 trajectories

WT

Tspan5

Tspan15 A

D

21 15

21 26

Fig. 6 Tspan5 and Tspan15 differentially affect ADAM10 dynamic behavior.aTIRF microscopy analysis of U2OS-N1 cells expressing or not GFP-tagged Tspan5 or Tspan15. Left column GFP signal;

middlelabeling of ADAM10, using a concentration of anti-ADAM10 Fab allowing single molecule detection. The images shown here are the first frame of the movies shown as supplementary information.

Right columnsuperimposition of 30 randomly selected ADAM10 tra- jectories with the DIC image of cells acquired before tracking.Bar 5lm. b Distribution of the apparent diffusion coefficients (ADC) calculated for all individual ADAM10 molecules analyzed in U2OS- N1 cells expressing or not Tspan5 or Tspan15. Eachdotrepresents one trajectory and 1500 trajectories are shown for each cell type. The

mean ADC value±SEM are indicated on theright.Triple asterisks indicate that the difference between the two cell types are significant with a p value below 0.001 as determined by the Mann–Whitney U test.c Distribution of the apparent diffusion coefficients (ADC) calculated for individual ADAM10 molecules according to their type of motion (Brownian, confined, mixed). Triple asterisks difference between the two cell types are significant with apvalue below 0.001 as determined by the Mann–WhitneyUtest.dHistogram representing the percentage of ADAM10 molecules undergoing Brownian, confined and mixed trajectories relative to the total number of trajectories, in U2OS-N1 cells expressing or not Tspan5 or Tspan15

(10)

0.067±0.027 (U2OS-N1) and 0.070±0.026 (Tspan5) (mean±SEM, n=1500,p\0.0001)] (Fig.6b). This is due to a higher ADC of molecules displaying a Brownian motion [0.138±0.030 versus 0.112±0.030 lm2/s (U2OS-N1) and 0.108±0.027 (Tspan5), n=1035, 795 and 840,p\0.0001] as well as to a lower percentage of molecules undergoing confined and mixed trajectories [16/

15 versus 26/21 % (U2OS-N1) and 23/21 % (Tspan5), Fig.6d], the diffusion of which is slower (Fig.6c). Alto- gether, these data show that the membrane environment of ADAM10 was modified upon Tspan15 expression, and indicate that the membrane compartmentalization of ADAM10 is differentially regulated by Tspan5 and Tspan15.

The repertoires of proteins co-immunoprecipitated with Tspan5 and Tspan15 display quantitative and qualitative differences

The above data suggest that Tspan5 and Tspan15 target ADAM10 to different membrane compartments. As a first step to identify these compartments, we compared by label- free quantitative mass-spectrometry the repertoire of plasma membrane-resident integral proteins co-immuno- precipitated with Tspan5, Tspan15 or CD9 (a classical tetraspanin) after Brij97 lysis, a detergent that preserves any number of interactions inside the tetraspanin web, including tetraspanin to tetraspanin interactions. Most previous analyses of tetraspanin-associated proteins were performed after immunoprecipitation using anti-tetraspanin antibodies. In the absence of good antibodies to Tspan5 or Tspan15, we had to rely on an alternative approach. To evaluate the suitability of GFP-trap pull down of GFP- tagged tetraspanins, we compared the pattern of cell membrane proteins co-immunoprecipitated with CD9 (from U2OS-N1 cells) using a CD9 mAb, or with CD9- GFP (from U2OS N1 cells stably expressing CD9-GFP) using GFP-trap beads. These two approaches yielded very similar sets of co- immunoprecipitated proteins (supple- mentary Table 1), indicating the appropriateness of the GFP-trap approach. It should be, however, pointed out that several tetraspanins (CD81 or CD82 for example) were not or less well detected using the GFP-trap approach. Of note, both Tspan14 and Tspan5 were detected in the CD9 immunoprecipitation performed with the CD9 mAb (sup- plementary Table 1).

Table1shows the most abundant integral proteins pre- sent at the plasma membrane co-immunoprecipitated with CD9, Tspan5 or Tspan15 using the GFP-trap approach.

Importantly, ADAM10 was by far the most abundant protein co-immunoprecipitated with Tspan5 and Tspan15, indicating that it is the major protein associating with these tetraspanins in U2OS-N1 cells. Tspan15 co-

immunoprecipitated at high levels a number of proteins that were not or poorly co-immunoprecipitated with the other two tetraspanins. In contrast, no proteins were specifically co-immunoprecipitated with Tspan5. Impor- tantly, a number of proteins known to associate (directly or indirectly) with CD9 were better co-immunoprecipitated with Tspan5 than with Tspan15 (Table1; supplementary Table 1; Fig. 7b). These proteins include the CD9 and CD81 partners CD9P-1/EWI-F and EWI-2, as well as the two subunits of the integrin a3b1, a CD151 partner.

Western-blot analysis of the immunoprecipitates confirmed that these molecules are better immunoprecipitated with Tspan5 than with Tspan15 (Fig.7c). This is consistent with the better immunoprecipitation of CD9 and CD151 with Tspan5 than with Tspan15. Tspan14 co-immunoprecipi- tated intermediate levels of these proteins. Among the proteins that were found to better co-immunoprecipitate with Tspan15, we validated the interaction with CD97 and IgSF3 by Western blot (Fig.7c). The interaction with other identified proteins, such as APMAP, could not be validated because we did not find antibodies suitable for Western- blot analysis. We also validated that Tspan5 and Tspan15 co-immunoprecipitated a similar level of the transferrin receptor and of the integrina2b1. The proportion of these receptors co-immunoprecipitated with tetraspanins was, however, lower than that of the other proteins tested.

Finally, among the ADAM10 substrates tested in this study, N-cadherin (cadherin-2) and CD44 were better co- immunoprecipitated with Tspan15 and Tspan5, respec- tively (supplementary Table 1).

Altogether, these data indicate that the repertoire of integral membrane proteins co-immunoprecipitated with Tspan5 more closely resembles that of CD9 than the repertoire of Tspan15-co-immunoprecipitated proteins.

Tspan5 and Tspan15 expressions have a different impact on the interaction of ADAM10 with other integral proteins

We then analyzed by quantitative mass-spectrometry the repertoire of membrane proteins associating with ADAM10 according to the expression of Tspan5 or Tspan15 (Table 2). Surprisingly, with the exception of chondroitin sulfate proteoglycan 4 and platelet-derived growth factor receptor beta, ADAM10 did not co-im- munoprecipitate the proteins specifically associated with Tspan15. In contrast, it co-immunoprecipitated a number of proteins associated with CD9 and Tspan5, including the integrin a3b1, CD9P1 and EWI-2. Importantly, these proteins were less efficiently co-immunoprecipitated with ADAM10 after Tspan15 expression than after Tspan5 expression. To confirm these findings and check that ADAM10 associated with these molecules at the cell

(11)

Table 1 Major plasma membrane associated integral proteins co-immunoprecipitated with CD9, Tspan5 or Tspan15

Protein name Gene name Protein abundance (area9106)

(number of unique peptides)

CD9 Tspan5 Tspan15

CD9-specific

Sodium/potassium-transporting ATPase subunit beta-1 ATP1B1 71.37 (3) ND ND

Monocarboxylate transporter 8 (MCT 8) SLC16A2 40.67 (2) ND ND

Tsp5&Tsp15

Disintegrin and metalloproteinase domain-containing protein 10 ADAM10 27.69 (5) 3795.00 (18) 4232.00 (31)

Transferrin receptor protein 1 TFRC 81.55 (23) 50.40 (16) 49.01 (19)

CD9 antigen CD9 7170.00 (5) 49.59a (1) 48.51a (1)

4F2 cell-surface antigen heavy chain SLC3A2 44.10 (16) 32.84 (9) 20.20 (6)

Sodium/potassium-transporting ATPase subunit alpha-1 ATP1A1 145.10 (25) 26.19 (15) 27.30 (13) Disintegrin and metalloproteinase domain-containing protein 17 ADAM17 45.01 (8) 24.14 (5) 24.56 (4) HLA class I histocompatibility antigen, B alpha chain HLA-B 14.22 (2) 23.77 (1) 34.52 (2)

Integrin alpha-2 ITGA2 7.76 (2) 21.53 (15) 15.90 (6)

Basal cell adhesion molecule BCAM 22.46 (12) 18.76 (6) 18.25 (3)

Teneurin-3 TENM3 61.92 (31) 18.54 (6) 24.03 (9)

Tetraspanin-14 TSPAN14 ND 4.42 (1) 5.09 (2)

Sodium/potassium-transporting ATPase subunit beta-3 ATP1B3 55.54 (3) 3.54a (1) 5.33a (1) Tsp5[Tsp15

Tetraspanin-5 TSPAN5 3.89 (1) 3270.00 (6) 7.80a (1)

Integrin beta-1 ITGB1 514.40 (16) 423.40 (17) 103.50 (9)

Integrin alpha-3 ITGA3 515.30 (23) 288.00 (17) 99.67 (14)

CD44 antigen CD44 89.57 (5) 28.47 (2) 14.25 (3)

Erythrocyte band 7 integral membrane protein STOM 9.02 (3) 28.43 (9) 2.90 (3)

Tetraspanin-9 TSPAN9 21.06 (2) 28.09 (2) 6.07 (2)

Prostaglandin F2 receptor negative regulator (CD9P-1/EWI-F) PTGFRN 808.80 (28) 26.32 (8) 7.42a (1)

CD151 antigen CD151 12.10 (1) 26.09 (1) ND

Syntaxin-4 STX4 79.67 (6) 21.36 (6) 9.50a (1)

CD166 antigen ALCAM 10.17 (4) 21.33 (1) 10.09 (2)

Low-density lipoprotein receptor LDLR ND 21.02 (1) 9.15a (3)

Cell surface glycoprotein MUC18 MCAM 4.52 (1) 20.11 (8) ND

Latrophilin-2 LPHN2 24.63 (12) 18.62 (2) ND

Integrin alpha-6 ITGA6 114.70 (27) 17.17 (9) 15.82a (6)

Immunoglobulin superfamily member 8 (EWI-2) IGSF8 218.00 (4) 14.66 (5) 6.87a (2)

Cytoskeleton-associated protein 4 CKAP4 ND 36.13 (11) 20.22 (4)

Tsp15[Tsp5

Platelet-derived growth factor receptor beta PDGFRB 1.98 (3) 6.70 (1) 87.70 (12)

CD97 antigen CD97 4.35 (2) 22.24 (5) 76.17 (9)

Matrix metalloproteinase-14 (MT1-MMP) MMP14 16.05 (4) 7.51a (1) 65.51 (3)

Syntaxin-3 STX3 40.02 (4) 19.91 (6) 54.13 (3)

Teneurin-2 TENM2 19.51 (17) 1.96 (1) 52.10 (2)

Sn1-specific diacylglycerol lipase beta DAGLB 3.19 (2) 7.59 (1) 47.95 (6)

Zinc transporter ZIP14 SLC39A14 14.56 (2) 6.39 (1) 27.73 (2)

Immunoglobulin superfamily member 3 IGSF3 1.74 (1) 5.27 (3) 24.10 (15)

Chondroitin sulfate proteoglycan 4 CSPG4 ND 5.07 (6) 43.96 (26)

Protein sidekick-2 SDK2 ND 13.11 (4) 36.49 (5)

C-type mannose receptor 2 MRC2 ND 12.50 (6) 33.23 (8)

Integral membrane protein 2C (Transmembrane protein BRI3) ITM2C ND 10.37 (2) 21.05 (1)

(12)

surface, ADAM10 was immunoprecipitated after biotin labeling and Brij97 lysis. The co-immunoprecipitated proteins were eluted in RIPA buffer and identified through a second immunoprecipitation step (Fig.8a). These data indicated that in the presence of Tspan15, ADAM10 co- immunoprecipitated a lower fraction of surface CD9, EWI- 2, CD151 and integrina3b1.

Finally, to validate that the reduced ability of ADAM10 to co-immunoprecipitate other cell membrane proteins in the presence of Tspan15 is not due to an impairment of the antibody to recognize a particular fraction of ADAM10, reciprocal experiments were performed. As shown in Fig.8b, CD9, CD9P-1, CD151 and the integrina3b1 co- immunoprecipitated a smaller fraction of ADAM10 in cells transfected with Tspan15 than in cells transfected with Tspan5 or Tspan14, or parental cells. In addition, these molecules co-immunoprecipitated Tspan5 or Tspan14, but little Tspan15.

Altogether these data indicate that Tspan15 inhibits to some extent the interaction of ADAM10 with classical constituents of the tetraspanin web.

Differential plasma membrane localization

of ADAM10 according to the expression of Tspan5 or Tspan15

The above data indicate that ADAM10 better interacts with classical components of the tetraspanin web when associ- ated with Tspan5 than when associated with Tspan15. We then studied whether these tetraspanins differentially reg- ulated the subcellular localization of ADAM10. Both

Tspan5 and Tspan15 were expressed at the plasma mem- brane (as also shown in Fig.1). Tspan15 was also enriched in an internal compartment that was identified as late endosomes as determined by its colocalization with CD63 (supplementary Figure 3). There was however no enrich- ment of ADAM10 in this compartment, as shown by the absence of detectable labeling of this compartment with an anti-ADAM10 mAb after permeabilization (supplementary Figure 3).

It had been previously demonstrated that several tetra- spanins were enriched at the periphery of breast cancer cells [37]. Similarly, we observed an enrichment at the periphery of U2OS-N1 cells, at the plane of cell attach- ment, of CD9 and other molecules of the tetraspanin web (CD9P-1 or CD81). There was also an enrichment of ADAM10 at the periphery of these cells, which was rein- forced when cells were incubated at 37C for 15 min with the anti-ADAM10 mAb 11G2 (supplementary Figure 4).

ADAM10 was also clearly enriched at the cell periphery in cells transfected with Tspan5, and as with parental U2OS-N1 cells, this peripheral labeling of ADAM10 was reinforced when cells were incubated at 37 C for 15 min with the anti-ADAM10 mAb (Fig.9). In contrast, there was no enrichment of ADAM10 at the cell periphery of U2OS-N1/Tspan15 cells, whether the cells were incubated with the anti-ADAM10 mAb or not.

This differential enrichment of ADAM10 at the cell periphery according to the expression of Tspan5 or Tspan15 led us to compare at the single molecule level the behavior of this protease at the cell center and at the cell periphery (supplementary Table 2). The dynamic behavior Table 1continued

Protein name Gene name Protein abundance (area9106)

(number of unique peptides)

CD9 Tspan5 Tspan15

Tetraspanin-15 TSPAN15 ND ND 6946.00 (8)

Adipocyte plasma membrane-associated protein APMAP ND NDb 274.40 (10)

Discoidin, CUB and LCCL domain-containing protein 2 DCBLD2 ND ND 38.92 (3)

Ephrin type-B receptor 2 EPHB2 ND ND 24.30 (8)

Latrophilin-1 LPHN1 ND ND 21.46 (10)

CD276 antigen CD276 29.80 (2) NDb 22.90 (3)

Dystroglycan DAG1 23.08 (6) ND 11.27a (2)

CD81 antigen CD81 14.00 (1) NDb 5.77 (1)

The table show the most abundant integral proteins known to be present at the plasma membrane and co-immunoprecipitated with GFP-tagged CD9, Tspan5 or Tspan15 using GFP trap beads. Only proteins identified in two experiments with an areaC29107in at least one of the IP are shown, except for tetraspanins. The values shown in this table are those obtained in the experiment performed using the highest number of cells.

The proteins in italic correspond to proteins for which the relative ratio in the Tspan5 and Tspan15 IP are different in the two experiments.

Proteins in bold character correspond to proteins not previously demonstrated to associate with tetraspanins. A complete list of proteins obtained in the two experiments is provided in supplementary table I

a Not detected in the second experiment in which less cells are used

b Detected in the second experiment

(13)

of ADAM10 in U2OS-N1/Tspan15 cells was similar at both locations. In contrast, the behavior of ADAM10 in U2OS-N1/Tspan5 cells differs according to the cell region analyzed, with Brownian trajectories being slower at the cell periphery than at the cell center.

Altogether, these data indicate that the ability of ADAM10 to localize into discrete membrane areas enri- ched in several tetraspanins and their partners is differentially regulated by its association with Tspan5 or Tspan15.

Discussion

Like other surface molecules associated with the tetra- spanin web, ADAM10 forms primary complexes with discrete tetraspanins. ADAM10 is, however, unusual by the fact that it associates directly with six tetraspanins, all

members of a particular subgroup characterized by eight cysteines in the large extracellular domain and referred to as TspanC8 [16, 18, 19]. We now demonstrate that the TspanC8 tetraspanins have a different impact on ADAM10 dependent Notch signaling and the cleavage of several ADAM10 substrates. This is associated with a different membrane compartmentalization of ADAM10.

TspanC8 tetraspanins have a different impact on Notch activity and the cleavage of ADAM10 substrates

The high number of tetraspanins that associate directly with ADAM10 suggests that each of these TspanC8 could confer to ADAM10 different properties. In this regard, two of the TspanC8 were shown to target ADAM10 to a late endosomal compartment in HeLa cells, whereas four of them (Tspan5, Tspan14, Tspan15 and Tspan33) increased

ADAM10 Tspan15-GFP Tspan5-GFP

U2OS-N1 cells transfected with :

Tspan5 Tspan15 -

(KDa)

130

55 170

70 100

35 45

25 15 IP GFP Trap

Coomassie staining

A

1/161/8 1/41/2 1 2 4

Ratio (log2 scale) Tspan5/Tspan15

(116)All Top 25 in CD9 IP

CD9-specific detected in Tspan5 but not Tspan15 IP

detected in Tspan15 but not Tspan5 IP detected in both Tspan5 and Tspan15 IP

IP GFP

(KDa) 55

Blot :

Cell lysate

CD9P1 130

ADAM10 70

EWI2 70

CD151

Integrin α3 170

CD9

Transfection

170 170 100 Tf. receptor

IgSF3 CD97 Integrin α2

BCAM 70

>170 35 20

CD81 20

C

B

Fig. 7 Analysis of the repertoire of proteins co-immunoprecipitated with Tspan5, Tspan15 or CD9.aThe proteins co-immunoprecipitated with Tspan5 or Tspan15 from Brij97 lysates of U2OS-N1/Tspan5 or U2OS-N1/Tspan15 using GFP Trap beads were separated by SDS/PAGE and visualized by Coomassie blue staining. As a control, the same procedure was applied to parental U2OS cells.b Distribution of the different integral plasma membrane proteins identified in CD9, Tspan5 or Tspan15 immunoprecipitates by quantitative mass-spectrometry, according to their relative abundance in the Tspan5 and Tspan15 immunoprecipitates. Eachdotcorresponds to a protein. Those proteins that are present in both immunoprecipitates are represented according to

their ratio in the Tspan5 IP versus the Tspan15 IP. The proteins immunoprecipitated with Tspan5 but not Tspan15, with Tspan15 but not Tspan5, and proteins only found in the CD9 IP are also shown. Theleft partof the graph shows all 116 identified proteins. Theright partshows the relative levels in the Tspan5 and Tspan15 IP of the 25 most abundant proteins in the CD9 IP. Note that most of them are more efficiently co- immunoprecipitated with Tspan5 than with Tspan15.cU2OS-N1 cells transfected or not with the indicated tetraspanins were lysed in Brij 97 before immunoprecipitation using GFP Trap beads. The composition of the immunoprecipitates was analyzed by western-blot

(14)

its surface expression levels [16]. Expression of Tspan5 and Tspan14 in HeLa cells stimulated ligand-induced Notch signaling, while silencing these two tetraspanins in U2OS-N1 cells reduced Notch signaling [16]. The effect of increasing or decreasing the expression of these TspanC8 on Notch activity correlated with a change in ADAM10 expression level. It was therefore not possible to determine whether the effect on Notch signaling was a mere effect of a change of ADAM10 surface expression levels, or whe- ther these tetraspanins also regulated the ability of ADAM10 to mediate Notch activation after ADAM10 has reached the plasma membrane. To address this question and test the effect of Tspan15 and Tspan33 on Notch

activation (U2OS-N1 cells express little amounts of these 2 TspanC8), we stably expressed Tspan5, Tspan14, Tspan15 and Tspan33 in these cells. Whereas all 4 TspanC8 induced an increase in ADAM10 surface expression levels, Tspan15 and Tspan33 expression reduced by *60 % ligand-induced Notch signaling. Expression of Tspan5 and Tspan14 did not result in an increase of Notch signaling, perhaps because this signaling is optimal in U2OS-N1 cells. The conclusion that Tspan5 and Tspan15 are, respectively, positive and negative regulators of Notch signaling is strengthened by the analysis of PC3 cells. We observed that Notch signaling in PC3 cells could be stim- ulated by OP9 cells expressing the Notch ligand DLL1.

Table 2 Plasma membrane-associated integral proteins co-immunoprecipitated with ADAM10 from U2OS-N1 cells expressing Tspan5 or Tspan15

Protein name Gene name Protein abundance (area/106)

(number of unique peptides)

Tspan5 Tspan15

Disintegrin and metalloproteinase domain-containing protein 10 ADAM10 372 (2) 662 (6)

Tetraspanin-5 TSPAN5 1370 (1) ND (5)

CD9 antigen CD9 346 (1) 178 (1)

Tetraspanin-14 TSPAN14 183 (1) 203 (2)

CD81 antigen CD81 51.2 (1) 25.9 (1)

Tetraspanin-9 TSPAN9 30.8 (1) 45.7 (1)

CD82 antigen CD82 3.87 (1) ND

CD63 antigen CD63 ND 5.96 (1)

Tetraspanin-15 TSPAN15 ND 3070 (5)

Integrin beta-1 ITGB1 684 (8) 256 (5)

Integrin alpha-3 ITGA3 984 (12) 262 (5)

Integrin alpha-6 ITGA6 132 (13) 33 (8)

Prostaglandin F2 receptor negative regulator (CD9P-1) PTGFRN 943 (17) 383 (11)

Immunoglobulin superfamily member 8 (IgSF8) (EWI-2) IGSF8 175 (7) 115 (5)

Choline transporter-like protein 1 (CTL-1) SLC44A1 44.6 (2) 31.8 (3)

Lactadherin short form MFGE8 34.7 (2) 14.1 (3)

Teneurin-3 TENM3 33.0 (3) 22.4 (1)

Adipocyte plasma membrane-associated protein APMAP 30.9 (1) ND

Transferrin receptor protein 1 TFRC 23.6 (6) 38.4 (9)

Syntaxin-4 STX4 22.1 (2) ND

Sodium/potassium-transporting ATPase subunit alpha-1 ATP1A1 18.1 (5) 80.6 (5)

Matrix metalloproteinase-14 (MT1-MMP) MMP14 8.64 (1) 54.0 (6)

Receptor-type tyrosine-protein phosphatase F PTPRF 8.22 (1) ND

4F2 cell-surface antigen heavy chain SLC3A2 7.66 (2) 8.44 (2)

Basal cell adhesion molecule (BCAM) BCAM 6.95 (2) ND

Zinc transporter ZIP14 SLC39A14 3.37 (1) ND

Chondroitin sulphate proteoglycan 4 CSPG4 ND 212 (28)

Platelet-derived growth factor receptor beta PDGFRB ND 4.69 (1)

The table shows the integral proteins known to be present at the plasma membrane and co-immunoprecipitated with ADAM10 from cells expressing Tspan5 or Tspan15. Only proteins identified with two unique peptides are considered, except for tetraspanins and proteins identified in the Tspan5 or Tspan15 immunoprecipitates. Note that the relatively low level of ADAM10 in the samples is due to the elution of co- immunoprecipitated material in Laemmli buffer at room temperature

(15)

This stimulation was abolished by treating the cells with an ADAM10 inhibitor and ac-secretase inhibitor, indicating the engagement of the canonical Notch pathway. We have previously shown that these cells express mainly Tspan15 and at a lower level Tspan5, as determined by RT-qPCR [16]. Importantly, silencing Tspan15 in these cells reduced ADAM10 expression level by 60 %, but increased Notch signaling twofold. In contrast, silencing Tspan5 reduced Notch signaling despite a lower impact on ADAM10

surface expression levels. Finally, we have shown that Tspan15 inhibited Notch signaling at a pre-c-secretase step. Altogether, these data indicate that ADAM10-de- pendent Notch signaling is facilitated by the presence of Tspan5 (and probably Tspan14) and inhibited by the presence of Tspan15 or Tspan33. It was previously demonstrated that the C. elegans TspanC8 tetraspanin, Tsp-12 and the three drosophila TspanC8 were positive regulators of Notch signaling [16, 17]. The finding that A

(KDa)

130

55 170

70 100

35 45

25 55 25

130

25

Transfection:

IP2:

(Strepta)

Elution

EWI-2 CD9 Int. α3 CD151 IP ADAM10

Strepta.

B

(KDa)

55 100

55 70

20 25

ADAM10

GFP CD9

Transfection

U2OS-N1 Tspan14 Tspan5 Tspan15

Blot: IP:

100 32 4 8 5 0 100 8 2 4 3 0 100 18 2 6 5 0 100 3 0.5 1 1 0

Fig. 8 Expression of Tspan15 weakens the association of ADAM10 with classical components of the tetraspanin web. a After biotin labeling of surface proteins, U2OS-N1 cells expressing Tspan5, Tspan14 or Tspan15 were lysed in the presence of Brij 97 and ADAM10 was immunoprecipitated using mAb 11G2-coupled Sepharose beads. The co-immunoprecipitated proteins were eluted in RIPA buffer, and selected proteins were immunoprecipitated using specific antibodies. The proteins were visualized by western-blotting

using fluorescent streptavidin. b U2OS-N1 cells expressing or not Tspan5, Tspan14 or Tspan15 were lysed in the presence of Brij 97 before immunoprecipitation as indicated. The composition of the immunoprecipitates was analyzed by Western blotting using a combination of biotin-labeled mAb and fluorescent streptavidin. A relative quantification of ADAM10 in the different immunoprecipi- tates in shown

Figure

Updating...

References

Related subjects :